What Is NeuroFinance?

One of the first ques­tions I am asked when­ev­er I present my research usu­al­ly is “what is neu­ro­fi­nance?”. Below is my view on this emerg­ing field of research, defin­ing it, espe­cial­ly with regards to behav­ioral finance and neu­roe­co­nom­ics.

The Neurofinance Paradigm

In neu­ro­fi­nance, we exam­ine exper­i­men­tal­ly the nature of the cog­ni­tive process­es engaged in acquir­ing and pro­cess­ing infor­ma­tion in finan­cial deci­sion mak­ing. We fur­ther study how peo­ple select action plans based on the acquired rep­re­sen­ta­tions of the val­ues of poten­tial invest­ment prospects.  One of our goals is to iden­ti­fy what kind of infor­ma­tion the human brain can process effi­cient­ly (and what kind it can­not), as well as the envi­ron­men­tal con­di­tions facil­i­tat­ing or ham­per­ing this infor­ma­tion pro­cess­ing. Anoth­er goal is to bet­ter under­stand how invest­ment deci­sions are tuned depend­ing on the appre­ci­a­tion of dis­tinct kinds of uncer­tain­ty, such as risk, jump risk, and esti­ma­tion uncer­tain­ty (ambi­gu­i­ty and mod­el uncer­tain­ty).

A new Kind of Behavioral Finance

Behav­ioral finance emerged in the 90s to per­fect the insights of math­e­mat­i­cal finance. The point of depar­ture of behav­ioral finance is that because clas­si­cal finance assumes full ratio­nal­i­ty, it can­not explain many price pat­terns. Using insights from all behav­ioral sci­ences (cog­ni­tive neu­ro­science, psy­chol­o­gy, soci­ol­o­gy) on how real peo­ple depart from the ratio­nal mod­el — real peo­ple are bound­ed­ly ratio­nal, behav­ioral finance can ratio­nal­ize hith­er­to-puz­zling price pat­terns.

The epis­te­mol­o­gy under­ly­ing neu­ro­fi­nance is dif­fer­ent and reflects recent advances in deci­sion neu­ro­science. We’re ini­tial­ly agnos­tic about the degree of ratio­nal­i­ty of peo­ple, i.e, we do not take peo­ple to be lim­it­ed in their com­pu­ta­tion­al capa­bil­i­ties. Rather, we infer their degree of sophis­ti­ca­tion exper­i­men­tal­ly, from the obser­va­tion of behav­ior and neur­al activ­i­ty dur­ing cog­ni­tive tasks per­formed in the lab. These cog­ni­tive tasks repli­cate chal­lenges that are rou­tinely encoun­tered in real world finan­cial deci­­sion-mak­ing.

These chal­lenges include:

  • Learn­ing asset dis­tri­b­u­tions that jump over time (objec­tive­ly very hard but in effect eas­i­er for peo­ple than one would thought; for more info, see this 1-page arti­cle in the uni­ver­si­ty jour­nal)
  • Learn­ing to avoid seem­ing­ly-glam­orous but sub­op­ti­mal invest­ments (not easy because we lack self-con­trol! For more info, see this work­ing paper)
  • Prop­er­ly per­ceiv­ing finan­cial mar­ket returns (not easy either, owing to some ingrained bias­es that plague human per­cep­tion!)
  • Mak­ing every­day pre­dic­tions about key finan­cial phe­nom­e­na such as price changes, etc. (under study!)

One impor­tant aspect of this new par­a­digm is to exam­ine in the lab which envi­ron­men­tal con­di­tions ham­per the emer­gence of ratio­nal­i­ty, and which con­di­tions help make peo­ple smart. Thus, Neu­ro­fi­nance affords a unique oppor­tu­ni­ty to

  • devel­op acute pre­dic­tion of investors’ behav­ior
  • iden­ti­fy envi­ron­men­tal mark­ers of behav­ioral sophistication/irrationality in finan­cial mar­kets
  • cre­ate nudges to aid deci­sion mak­ing

Why looking at Neural Activity?

Method­ol­o­gy-wise, neu­ro­fi­nance lies at the inter­sec­tion of exper­i­men­tal eco­nom­ics and com­pu­ta­tion­al neu­ro­science. We repli­cate in the lab core chal­lenges faced by finance prac­ti­tion­ers, and we exam­ine how lab sub­jects (reg­u­lar peo­ple as well as finance pro­fes­sion­als) solve these chal­lenges.

The ques­tion is: Do the cog­ni­tive process­es that the sub­jects imple­ment approx­i­mate the opti­mal solu­tion, which “Mr Spock” (the ratio­nal agent) would imple­ment? Or are these cog­ni­tive process­es more akin to the bound­ed­ly ratio­nal heuris­tics which “Homer Simp­son” would use1?

To answer this ques­tion, we do two things:

  • Look at behav­ior: Some­times, from observ­ing the choic­es of a sub­ject through­out the exper­i­ment, we can infer to what extent the sub­ject act­ed more like Mr Spock, or more like Homer. This kind of infer­ence works well when Mr Spock and Homer would behave dif­fer­ent­ly in the task, which is often the case.
  • Scan the brain of the sub­jects dur­ing the exper­i­ment: If we iden­ti­fy brain regions with a response pro­file con­sis­tent with the spe­cif­ic com­pu­ta­tion­al process per­formed by Mr Spock (resp Homer), the behav­ioral evi­dence that sub­jects act­ed more like Mr Spock (resp Homer) is strength­ened.

One exam­ple: To learn opti­mal­ly the expect­ed returns of assets that jump over time, investors must acquire Bayesian jump detec­tion sig­nals, which they use at each point in time to tune their learn­ing rate. The plau­si­ble alter­na­tive to this Bayesian learn­ing, rein­force­ment learn­ing, does noth­ing of the kind. So, iden­ti­fy­ing brain regions whose activ­i­ty cor­re­lates with the Bayesian sig­nals enables the infer­ence that sub­jects approx­i­mat­ed Bayesian learn­ing. The infer­ence is pow­er­ful because it is very unlike­ly that the iden­ti­fi­ca­tion of these neur­al sig­nals be the result of serendip­i­ty.

Computational neuroeconomics Applied to Finance

This com­pu­ta­tion­al approach reflects a new trend in neu­roe­co­nom­ics. By iden­ti­fy­ing regions that imple­ment a spe­cif­ic com­pu­ta­tion­al process, instead of mere­ly report­ing the “acti­va­tion” of a brain region in a giv­en exper­i­men­tal con­di­tion (which involves many com­pu­ta­tion­al process­es), this approach enables a more con­vinc­ing form of infer­ence than is tra­di­tion­al­ly made in func­tion­al imag­ing stud­ies.

What for? Implications for the industry

Port­fo­lio man­agers and traders have to process infor­ma­tion on the spot in rapid­ly chang­ing envi­ron­ments. Lit­tle is known about how to tai­lor orga­ni­za­tion­al and indi­vid­ual deci­sion-mak­ing process­es to help peo­ple process infor­ma­tion effi­cient­ly in such con­texts. By iden­ti­fy­ing envi­ron­men­tal fac­tors improv­ing effi­cient infor­ma­tion pro­cess­ing, it is hoped that research in neu­ro­fi­nance will pro­duce prac­ti­cal results on how to improve invest­ment and trad­ing deci­sions, at both indi­vid­ual and orga­ni­za­tion­al lev­els.

  1. You will have rec­og­nized the fig­ures employed by Richard Thaler to illus­trate these dif­fer­ent meth­ods in his book Nudge []